Processing and data storage happen on edge systems over the cloud. But network constraints may be the best way to distinguish edge from cloud.
At the core, the key difference between edge computing and cloud computing comes down to one concept: network connection constraints. If internet connectivity were to be continuously available everywhere and data could be transferred immediately without any delays, there would be no need for edge computing solutions. But, since network connections are neither constant nor omnipresent, and both bandwidth and latency are limited, edge computing solutions have emerged to address each of these issues.
With cloud computing, an internet connection lets users leverage processing power and data remotely. For most purposes, users can act as if the processing power and storage is infinite, even if computing capacity is actually constrained by a finite number of data centers. Need more processing power? Spin up more servers. Prefer to preserve more photos, files or other data? Provision larger storage limits.
The rise of both reliable internet connections and cloud computing occurred roughly in sync from the early 2000s on. The type of data transmitted changed over time, too, gradually evolving from a focus on text and compressed images to audio and livestreamed video.
For organizations, this meant formerly on-site file servers might be moved to the cloud. Cloud computing companies often offered reliability and redundancy levels that easily exceeded what most I.T. departments could provide.
Cloud applications brought both power and management simplicity to more organizations. The responsibility for updates, for example, shifted from staff administrators to cloud vendors. For example, new features appear in Google Docs when Google rolls out a change, no action needed from an on-site administrator.
Many school systems, chronically resource-constrained, switched to Chromebooks since the cloud-centric computers are simpler to manage, maintain and secure than most legacy server-centric systems.
SEE: Don’t curb your enthusiasm: Trends and challenges in edge computing (TechRepublic)
Two niche segments—large interactive website providers and streaming services—provided the first hints that a distinctly different computing architecture might be needed. Specifically, these firms realized that data delivery to individual customer computers takes time. Content delivery network providers began to move data centers closer to consumers (Figure A).
A person streaming a movie in Michigan, for example, will likely experience lower latency when the streaming source is a data center in Chicago versus Los Angeles. Today, vendors apply the term edge to various sizes of smaller data centers, but all of these seek to solve the network connection constraint of latency caused by distance.
Another class of edge computing emerged as connected home technologies were deployed. Users want smart locks, security cameras and various environmental sensors to work even when Wi-Fi goes down, a cable gets cut or cell networks temporarily stop working.
Eventually, vendors began to build systems that communicate using various standards, such as Bluetooth, Zigbee and Matter. Similarly, in an industrial environment, machines may need to operate in environments without reliable networks. Since nearly all of these smart home and industrial devices generally remain stationed in a single site, the network connection constraint they address is a temporary lack of connection.
Moving vehicles—on land, air and sea—present many of the most challenging edge computing tasks to date, though. Not only do automated cars, planes and ships need to operate in places without consistent internet connections (Figure B), but their rate of motion often leaves insufficient time to rely on remote calculations, even when connected. In other words, by the time a speed autonomous car might receive a reply from a nearby cloud data center about a road hazard, the car may already have met it.
Periodic connectivity for these sorts of vehicles remains critical. Cars and trucks, for example, benefit from updated maps to indicate roads, road work and current traffic conditions. Drones and autonomous ships rely on weather data to identify potentially turbulent skies or seas.
People managing these devices will likely want to offload data, not only to assess and enhance performance but also to retrieve images and other information acquired by the vehicle’s various instruments. Edge systems that allow for this sort of autonomous operation and intermittent network connection present several sets of challenges that, as of late 2022, are not yet fully solved.
What’s your experience?
Cloud computing makes it possible to leverage computing power and storage anywhere there is a reliable internet connection. Edge computing shifts processing and storage away from these cloud data centers to devices capable of computing independent of a constant connection.
Do you rely on a mix of cloud and edge computing systems in your work? Are there specific edge computing capabilities not mentioned above that you find helpful in your environment? What sorts of administrative and management systems have you deployed to help manage your organization’s cloud and edge computing systems?
Mention or message me on Twitter (@awolber) to let me know what your experience with edge computing has been—and which edge computing technologies you are watching most closely.